(Back to users list)
Op 01-08-2022 om 11:24 schreef Sven Barczyk:
Will the reboot of nodes force the virsh pools to be updated ?
Yes. Libvirt keeps the pools in memory. So what technically also works is:
1. Migrate all VMs away
2. Stop CloudStack agent
3. Stop Libvirt
4. Start Libvirt
5. Start CS Agent
Technically that's sufficient, but I suggest a reboot just to make sure.
Wido
restarting cloudstack-agent, did not do the magic, so is there an process
which is only triggered by restarting a node ?
kind regards
Sven
-----Ursprüngliche Nachricht-----
Von: Wido den Hollander <[email protected]>
Gesendet: Montag, 1. August 2022 11:12
An: [email protected]; Nazmul Parvej <[email protected]>
Cc: Sven Barczyk <[email protected]>; Product Development | BEXIMCO IT
<[email protected]>
Betreff: Re: Changing the connecting hosts when using Ceph as primary
Op 31-07-2022 om 14:19 schreef Nazmul Parvej:
Hi Sven and Wido,
I am facing the same issue with my ACSv4.17. I added my RBD Primary
Storage with Single RADOS monitor IP, Now I want to change it to a
round-robin DNS following are my SoP please correct me if I am wrong.
*1. Change in MySQL database: *There are two tables in the cloud
database called storage_pool and storage_pool_view I will update both
of them to my round-robin DNS.
The _view table is a SQL view. You only need to change the storage_pool table.
*2. Apply the Impact to Hosts:* One by one host restart to reload
Libvirt and the CloudStack agent to the right XML loaded.
Correct. A reboot is the easiest way.
Wido
Yours sincerely,
Nazmul Parvej
On Fri, Jul 29, 2022 at 7:56 PM Wido den Hollander <[email protected]> wrote:
Hi,
The easiest way is to reboot your hosts one by one. This will reload
Libvirt and the CloudStack agent and will make sure you have the
right XML loaded.
Wido
On 7/27/22 10:22, Sven Barczyk wrote:
Hello everyone,
i caught myself beeing stupid.
While deploying my cluster, I’ve decided to use my mon1.ceph to be
the main RBD Host which leads now to the problem, that a restart of
mon1 will be hell.
In despite, i changed the host to a round-robin dns (changed by sql
), which is the suggested method of the documentation and thought
all new Instances will be deployed with this new host from the database.
But no , still uses the mon1.ceph while deploying new instances
(checked with virsh dumpxml on my hosts), instead of the new set
host, mons.ceph which round-robins through all my mons.
Restart of Cloudstack-management did not help.
Is there anyone who could give me an hint where this rbd hosts might
be hardcoded in the cloudstack-config ?
Best regards,
Sven
